Configuration Guide

10.4

Back Next

Match/Search Strategy

For fuzzy-match base objects, the match/search strategy defines the strategy that the match process uses for searching and matching records. The match/search strategy determines how to match candidate A with candidate B by using fuzzy or exact options.

For more information about fuzzy-match base objects, see Match/Search Strategy.

The match/search strategy can affect the quantity and quality of the match candidates. An exact strategy requires clean and complete data; if the data is not cleansed or if the data is incomplete, the match process might miss some matches. A fuzzy strategy finds many more matches, but many might not be duplicates. When defining match rule properties, you must find the optimal balance between finding all possible candidates and avoiding irrelevant candidates.

Select one of the following strategy options:

Match/Search Strategy Option	Description
Fuzzy strategy	A probabilistic match that takes into account spelling variations, possible misspellings, and other differences.
Exact strategy	An exact match that matches records that are identical.

All fuzzy-match base objects have a fuzzy match key, which is generated based on the columns that you specify in the match path component. When you use the fuzzy strategy, the match process searches the match key index to identify match candidates. To ensure that the match process can identify candidates, you must specify at least one of the columns from the fuzzy match key in the match rule. For example, if the match key is generated based on the Name column, then the match rule must include the Name column.

The following table lists the types of match rules, the match/search strategies, and a description for each combination:

Match Rule Type	Match/Search Strategy Used	Description
Fuzzy-match rule	Fuzzy strategy	Contains fuzzy-match columns and might contain exact-match columns. Fuzzy-match rules are recommended for columns that contain data variations, such as misspelled names.
Exact-match rule	Exact strategy	Contains only exact-match columns. The match process identifies matches by running SQL statements on the rows in the database. Exact-match rules are best for columns, such as ID and date of birth, where you do not need fuzzy matching capabilities.
Filtered-match rule	Fuzzy strategy	Contains only exact-match columns. The rule runs on match candidates that are filtered based on the associated match key. The match process identifies match candidates by searching the match key index. It is recommended to use a filtered-match rule when one of the exact-match columns has the same source columns as the fuzzy match key. Filtered-match rules improve performance by running the exact-match rules in the Process Server instead of the database.

If you are constrained by performance issues related to the database server, consider using filtered-match rules instead of exact-match rules. Filtered-match rules let you run batches larger than what you can run on the exact-match rules. Also, for a large increment in batch size on filtered matching, the duration of the match job comparatively increases by a small margin.

Rename Saved Search

Table of Contents

Configuration Guide

Configuration Guide

Match/Search Strategy

Match/Search Strategy